International Journal of Environmental Research and Public Health

Article Bake-Out Strategy Considering Energy Consumption for Improvement of in Floor Heating Environments

Seonghyun Park 1 and Janghoo Seo 2,*

1 Department of Architecture, Graduated school, Kookmin University, 77 Jeongneung-ro, Seongbuk-gu, Seoul 02707, Korea; [email protected] 2 School of Architure, Kookmin University, 77 Jeongneung-ro, Seongbuk-gu, Seoul 02707, Korea * Correspondence: [email protected]; Tel.: +82-2-910-4593

 Received: 31 October 2018; Accepted: 30 November 2018; Published: 3 December 2018 

Abstract: Improved quality of life has led to a growing demand for better indoor air quality (IAQ). Buildings are becoming more airtight and insulated in order to minimize energy consumption. The importance of both energy conservation and IAQ improvement has been recognized and addressed by many studies. Bake-out is the process of using indoor heating to remove volatile compounds present in building materials and furnishings so that they can be vented out into the atmosphere. Indiscriminate use of heating to increase the surface temperature of materials during this process can result in significant loss of energy. Therefore, energy-efficient bake-out should be performed by considering both the floor temperature and the emission amount of pollutants. This study aims to investigate an effective and economical bake-out implementation strategy via experimentation and computational fluid dynamics analysis. The results showed weak direct correlation between the heating energy consumption and the amount of pollutants emitted. The study also highlights the passive option of installing sorptive building materials for improving IAQ economically.

Keywords: bake-out; indoor air quality; computational fluid dynamics; toluene; floor

1. Introduction Apartment houses and other newly constructed buildings are ensuring increased airtightness and insulation in their structures by using building materials made of complex chemicals in order to reduce energy consumption. However, volatile organic compounds (VOCs) such as formaldehyde (HCHO) and toluene emitted from building materials composed of complex chemicals degrade the indoor air quality. Workers or occupants continuously exposed to these environments, can contract lung cancer or respiratory diseases [1–4]. Therefore, it is necessary to make efforts to improve indoor air quality (IAQ). The methods for IAQ enhancement can be categorized into elimination, source removal and . Bake-out, which is one of the elimination methods, has drawn considerable attention and it is widely used to reduce indoor air pollutant concentration before occupants move into newly built residences. Much research has been conducted on bake-out and is still ongoing [5,6]. Lu et al. [7] conducted an empirical study on VOC elimination technology employing the chamber method for households in China and noted that bake-out performance is greatly influenced by the temperature and ventilation period. In addition, their study presents an ideal ventilation frequency for the optimal elimination of VOCs. Edwards et al. [8] studied the health effects of indoor air condition on occupants by taking samples of 30 VOCs in the air and analyzing the similarities between their pollutants. Kang et al. [9] studied the bake-out effect of floor heating systems in South Korean

Int. J. Environ. Res. Public Health 2018, 15, 2720; doi:10.3390/ijerph15122720 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2018, 15, 2720 2 of 16 households. Small-scale chamber experiments were conducted to gauge the amount of VOC emissions and a quantitative result was derived showing that shorter bake-out periods correspond to lesser effectiveness in reducing emissions. Lv et al. [10] investigated the effects of bake-out with dilution ventilation technology using experimental and numerical methods. These effects are influenced by bake-out temperatures, times and ventilated times. Lee et al. [11] presented an analytical model for the sinking behaviors of VOCs against porous building materials. Jiang et al. [12] investigated the emission characteristics of HCHO and VOCs using a particle board with high VOC emission in sealed and ventilated environmental chambers. The existing studies place much emphasis on identifying the emission characteristics of VOCs for different materials through surveys and experiments and implementing bake-out to remove these pollutants for better IAQ. It can be inferred from most of these findings that the elimination of VOCs from building materials can be accelerated by increasing the indoor temperature and the amount of ventilation using mechanical devices. However, these studies tend to focus only on enhancing the bake-out performance for reducing pollutant concentration. Although these methods may be effective for IAQ improvement, they could result in financial losses if too much energy is used in the process. Sorptive building materials (SBM), which are considered a passive method for pollutant adsorption, are an efficient tool that can reduce VOC concentration while replacing the existing building material at the same time. These materials remove chemical substances from the indoor air by physical sorption or chemical reaction [13]. Seo et al. [14] evaluated the effects of SBM on the reduction of VOCs through an experiment using a small chamber. Park et al. [15] explored effective installation methods of SBM in office environments by measuring and analyzing their VOC reduction effect per installation area based on occupant satisfaction. These methods reflect the need to develop an efficient and economical bake-out strategy for IAQ improvement by comparatively analyzing the cost of energy used during bake-out periods and reduction in VOC concentration. In this study, test chamber experiments and computational fluid dynamics (CFD) analysis are conducted by breaking down the bake-out procedure into two steps and measuring the changes in VOC concentration under different variables for each step. This study also analyzes the pollutant reduction efficiency in terms of the amount of energy used and reexamines the economic feasibility and practicality of the existing bake-out strategy, which focuses on performance.

2. Materials and Methods This study divides the bake-out procedure into two parts and conducts experimentation and CFD analysis. In the first step, the VOC concentration is measured using different variables in sealed mock-up chambers with an insulated floor, simulating the “bake” portion of the process. In the second step, the effectiveness of natural ventilation, mechanical ventilation and sorptive building materials are analyzed using CFD analysis, representing the “bake-out” portion. Table1 shows the heating and bake-out process conditions for each case. The floor temperature was set to 30, 40 and 50 ◦C and the VOC concentration was analyzed for each instance. For all cases, 40 mL of pollutant was injected. To evaluate the performance of pollutant concentration reduction by bake-out, three alternatives (ALTs) were set. The bake-out process is divided into natural ventilation, mechanical ventilation and installation of SBM.

Table 1. Floor temperature and bake-out conditions per case for bake-out experiment.

Case Set Floor Temp. Bake Process (Method) Bake-out Process (Method) Case 1 30 ◦C Floor heating (Experimental) N/A Case 2 40 ◦C ALT 1 Natural ventilation (CFD analysis) Case 3 50 ◦C Floor heating (Experimental and CFD analysis) ALT 2 Mechanical ventilation (CFD analysis) ALT 3 Installation of SBM (CFD analysis) CFD: computational fluid dynamics; ALT: alternatives; SBM: sorptive building materials; N/A: not applicable. Int. J. Environ. Res. Public Health 2018, 15, x 3 of 16

2.1.1. Target Space and Experiment Method Figure 1 illustrates an overview of the experiment. The experimental apparatus consists of the

Int.test J. chamber, Environ. Res. a Public pollutant Health 2018 detection, 15, 2720 device, a floor temperature control device and a temperature 3 of 16 sensor. The test chamber is a cube measuring 1800 mm on each side. It was built using Isopink for purposes and the edges were sealed airtight with silicon. The VOC detector was 2.1.filledInt. J. Experimental Environ. manually, Res. Public Conditionswhich Health means 2018, 15 that, x a sampling device was required to measure the level of 3VOC of 16 concentration at the center of the chamber. Stainless Steel (SUS 316, KWANGSHIN I.S.T Inc., 2.1.1.2.1.1. Target SpaceSpace andand ExperimentExperiment MethodMethod Hamyang, Korea), which has no VOC emission, was used for the construction of the sample injectionFigureFigure tube1 1 illustrates illustratesand the piping an an overview overview union of andof the the a experiment. 100experime mL medicalnt. TheThe experimental syringeexperimental was used apparatusapparatus as an consistsairconsists pump. ofof theThethe testmeasurementtest chamber,chamber, a intervala pollutant pollutant was detection detectionset to 10 device, min. device, a floor a floor temperature temperature control control device device and a temperatureand a temperature sensor. Thesensor. test The chamber test chamber is a cube is measuring a cube measuring 1800 mm 1800 on eachmm side.on each It was side. built It was using built Isopink using forIsopink thermal for insulationthermal insulation purposes purposes and the and edges the were edges sealed were airtightsealed airtight with silicon. with silicon. The VOC The detectorVOC detector was filled was manually,filled manually, which meanswhich thatmeans a sampling that a sampling device was device required was to required measure theto measure level of VOC the concentrationlevel of VOC atconcentration the center of at the the chamber. center of Stainless the chamber. Steel (SUS Stainless 316, KWANGSHIN Steel (SUS 316, I.S.T KWANGSHIN Inc., Hamyang, I.S.T Korea), Inc., whichHamyang, has no Korea), VOC emission, which has was no used VOC for theemission, construction was used of the for sample the injectionconstruction tube andof the the sample piping unioninjection and tube a 100 and mL the medical piping syringe union wasand useda 100 as mL an medical air pump. syringe The measurement was used as intervalan air pump. was set The to 10measurement min. interval was set to 10 min.

Figure 1. Diagram of the Experiment.

Figure 2 shows the temperature sensor locations within the chamber. A total of 11 sensors were used to ensure uniform temperature control: four on the floor (F_1–4) and one at the center of each side wall as well as the ceiling. In addition, there was a concentration sampler at the center of the chamber. Table 2 shows the details of the experimental equipment used for the experiment in this study. A positive temperature coefficientFigureFigure (PTC) 1.1. film Diagram was usof edthe to Experiment. simulate floor heating in order to produce an even control of temperature and the heating was supplied electrically. This made it possible to calculateFigureFigure the2 2 showsamount shows the the of temperature energytemperature used sensorfor sensor temper locations locationsature withincontrol within the andthe chamber. chamber.maintenance A A total total for of ofeach 11 11 sensors case.sensors were were usedusedTo toto ensuresimulateensure uniform uniform the presence temperature temperature of pollutants, control: control: four a four total on on theof the 40 floor floorg of (F_1–4) adhesives(F_1–4) and and one (two one at theLoctite at centerthe 401),center of eachwhich of each side is wallhighlyside aswall concentrated well as aswell the as ceiling. inthe VOC, ceiling. In addition,was In sprayed addition, there on was therethe a floor concentration was and a concentration allowed sampler to dissipate. sampler at the center at the of center the chamber. of the chamber. Table 2 shows the details of the experimental equipment used for the experiment in this study. A positive temperature coefficient (PTC) film was used to simulate floor heating in order to produce an even control of temperature and the heating was supplied electrically. This made it possible to calculate the amount of energy used for temperature control and maintenance for each case. To simulate the presence of pollutants, a total of 40 g of adhesives (two Loctite 401), which is highly concentrated in VOC, was sprayed on the floor and allowed to dissipate.

Figure 2. Sensor locations within the chamber. Figure 2. Sensor locations within the chamber. Table2 shows the details of the experimental equipment used for the experiment in this study.

A positive temperature coefficient (PTC) film was used to simulate floor heating in order to produce an even control of temperature and the heating was supplied electrically. This made it possible to calculate the amount of energy used for temperature control and maintenance for each case.

Figure 2. Sensor locations within the chamber.

Int. J. Environ. Res. Public Health 2018, 15, 2720 4 of 16

Table 2. Details of the experimental apparatus.

Component Parameter Description Room Condition Cooling temp. 25 ◦C Ceiling height 1800 mm Test chamber Floor area 1800 mm × 1800 mm Material Isopink (30 T) Temp. sensor Thermocouple (T-type) Monitoring Data Logger (GL 220) Heater Heating film (PTC) Temp. control PID controller (NX_1) Power consumption Power manager (B200) Experimental Equipment Pollutant Loctite 401 (20 g) Monitoring Three notebooks Sampling tube Sampler Sterile syringe (100 mL) 1005TLL 5.0 mL syringe

To simulate the presence of pollutants, a total of 40 g of adhesives (two Loctite 401), which is highly concentrated in VOC, was sprayed on the floor and allowed to dissipate. Table3 lists the specifications of the VOC analyzer (SGVA-P2, FIS Inc., Itami, Japan) used in this study. This VOC analyzer is a gas-chromatography-type gas-concentration-measuring device that uses a metal–oxide–semiconductor gas sensor and employs high-purity air as the carrier gas. The compounds separated though the columns are detected using a semiconductor gas sensor. Finally, the concentrations of the target gas are automatically calculated and indicated. The correlation of the results obtained from this analyzer with those obtained from the gas chromatography/mass spectrometry (GC/MS) method was more than 0.99 [16–18]. Samples for the VOC analysis were obtained every 10 min, considering the performance of the VOC analyzer (8 min measurement time). As there was a difference in the distance between the center of the chamber and the sampling point, the volume of air corresponding to the internal volume of the sampling tube was collected for the first time using a sterile syringe. Thereafter, a sampling syringe was used to obtain a 5 cc sample for the second time and to insert it into the VOC analyzer.

Table 3. Specifications of volatile organic compounds (VOC) analyzer (SGVA-P2).

Parameter Description Measurement principle Gas chromatography using semiconductor gas sensor Carrier gas High purity cylinder air Sampling method Manual sampling with a syringe (5 cc) Decomposition capability 0.1 ppb Measurement time 8 min Power supply 100–240 V, 50/60 Hz toluene ethylbenzene m-xylene styrene Measurement range 5–1000 ppb 5–1000 ppb 5–1000 ppb 5–1000 ppb

The experiment lasted for 10 h and included an indoor cooling period, temperature stabilization and pollutant measurement. To fully eliminate residual pollutants within the chamber after the experiment of each case, experiments were conducted at intervals of one week per case (total of three weeks) and the background concentration before measurement was measured.

2.1.2. Floor Temperature Control The energy used for heating must be quantitatively derived in order to analyze the effectiveness of baking at varying degrees of energy consumption. This study employed a PTC heating film to Int. J. Environ. Res. Public Health 2018, 15, 2720 5 of 16 obtain an even distribution of heat across the floor. The floor temperature was controlled using the PID control method, which can deliver timely response to the input values and has a small offset through the use of proportional, integral and differential parameters. Its basic formula is given in EquationInt. J. Environ. (1). Res. In Public addition, Health the 2018 input, 15, x variables of proportional (P), integral (I) and differential (D) for5 of the 16 PID controller were set to 5%, 60 s and 15 s, respectively.  1 1 Z de t de(t)  β(t) = Kp et + etdt + Td (1) β(t) = Kp e(tT) + e(t)dt +dtTd (1) i Ti dt

where β is the output value, e is the error factor, Kp is the proportional gain, Ti is the integral time and where β is the output value, e is the error factor, Kp is the proportional gain, Ti is the integral time and Td is the differential time. Td is the differential time. A T-type thermocouple was used to measure the temperature while the Data Logger (GL 220, A T-type thermocouple was used to measure the temperature while the Data Logger (GL 220, GRAPHTEC Co., Yokohama, Japan) kept track of the data and monitored the values. The GRAPHTEC Co., Yokohama, Japan) kept track of the data and monitored the values. The instantaneous instantaneous and total energy consumption of the system were measured using a Power Manager and total energy consumption of the system were measured using a Power Manager (B200S, (B200S, DAWONDNS Co., Gwangju, Korea). The room outside the chamber was cooled to a set DAWONDNS Co., Gwangju, Korea). The room outside the chamber was cooled to a set temperature temperature of 25 °C for 30 min in order to keep the temperature outside the chamber static. Then, a of 25 ◦C for 30 min in order to keep the temperature outside the chamber static. Then, a heating source heating source was used to measure the temperature change and energy consumption. The changes was used to measure the temperature change and energy consumption. The changes in temperature in temperature were measured in intervals of 10 s. were measured in intervals of 10 s.

2.2.2.2. NumericalNumerical ModelModel ofof CFDCFD AnalysisAnalysis

2.2.1.2.2.1. BoundaryBoundary ConditionCondition ofof CFDCFD AnalysisAnalysis ThisThis studystudy aimsaims toto analyzeanalyze thethe pollutantpollutant concentrationconcentration usingusing CFDCFD analysisanalysis andand developdevelop anan efficientefficient bake-outbake-out strategy. strategy. The The concentration concentration of VOC of emissionsVOC emissions from building from building materials ormaterials adhesives or isadhesives influenced is internallyinfluenced by internally the physical by the and physical chemical and properties chemical of properties the material of andthe externallymaterial and by temperature,externally by air temperature, velocity, turbulence air velocity, intensity turbulence and relative intensity . and relative This calls humidity. for a sufficient This calls amount for a ofsufficient information amount regarding of information these influencing regarding these factors. influencing However, factors. the existing However, method the existing for measuring method VOCfor measuring concentration VOC relies concentration on a single relies variable on a and single only variable assesses and how only an individualassesses how factor an affectsindividual the ventilationfactor affects characteristics the ventilation [19 ].characteristics Numerical analysis [19]. Numerical of CFD allows analysis the assessmentof CFD allows of pollutant the assessment emission of inpollutant the air underemission different in the variablesair under using different various variables models. using various models. FigureFigure3 3 shows shows the the target target space space and and mesh mesh for for the the CFD CFD analysis. analysis. This This space space was was assumed assumed to to have have thethe samesame sizesize as the experimental experimental space, space, with with an an inlet inlet and and an an outlet outlet set set on on the the sides sides for for ventilation. ventilation. In Inaddition, addition, the the SBM SBM was was assumed assumed to tobe be install installeded on on the the entire entire surface surface of ofthe the ceiling. ceiling.

(a) (b)

FigureFigure 3.3. ModelModel usedused forfor computationalcomputational fluidfluid dynamicsdynamics (CFD)(CFD) analysis:analysis: ((aa)) Geometry;Geometry; ((bb)) MeshMesh (Section(Section ABCD).ABCD).

Table 4 shows the boundary conditions for the CFD analysis used in this study. ANSYS Workbench 17.2 and FLUENT CFD engines were used for the analysis [20]. There were approximately 400,000 mesh elements and a no-slip condition finish was applied to the sidewalls of the test chamber. The dimensions of the space, the temperature of the test chamber walls, the heated temperature of the floor and the pollutant set values were set according to the experiment. All Int. J. Environ. Res. Public Health 2018, 15, 2720 6 of 16

Int.Int. J. J.Environ.Table Environ.4 Res.shows Res. Public Public the Health Health boundary 2018 2018, 15, 15 conditions, x, x for the CFD analysis used in this study. ANSYS Workbench6 6of of 17 17 17.2 and FLUENT CFD engines were used for the analysis [20]. There were approximately 400,000 mesh Int.interpretationselementsInt.Int.interpretations J. J. J.Environ. Environ. Environ. and Res. Res. Res. a Publicno-slip Public werePublicwere Health Health Health condition performedperformed 2018 2018 2018, ,15 ,15 15, finish ,x ,x x inin was anan applied unsteadyunsteady to the statestate sidewalls toto observeobserve of the test thethe chamber. changeschanges The inin dimensions pollutantpollutant66 6 of ofof 17 1717 concentrationofconcentration the space, the over over temperature time. time. of the test chamber walls, the heated temperature of the floor and the interpretations were performed in an unsteady state to observe the changes in pollutant interpretationspollutantinterpretations set values werewere were performedperformed set according inin an toan theunsteadyunsteady experiment. statestate All toto interpretations observeobserve thethe changes werechanges performed inin pollutantpollutant in an concentration over time. concentrationunsteadyconcentrationTable stateTable over 4. toover 4. Boundaryobserve Boundary time. time. thecondition condition changes of of CFD in CFD pollutant (computational (computational concentration fluid ) dynamics) over time. analysis analysis for for bake-out. bake-out.

ClassificationClassificationTable 4. Boundary condition ALT ALT of 1CFD 1 (computational fluid ALT ALT dynamics) 2 2 analysis for bake-out. ALT ALT 3 3 TableTableTable 4. 4. BoundaryBoundary Boundary condition condition condition of of of CFD CFD CFD (computational (computational fluid fluid fluid dynamics) dynamics) analysis analysis for for bake-out. bake-out. ClassificationClassificationClassification ALT ALT ALT 1 1 1 ALT ALT ALT 2 2 2 ALT ALT ALT 3 3 3 Classification ALT 1 ALT 2 ALT 3 Natural ventilation Mechanical ventilation Installation of SBM

Bake-outBake-out method method

Bake-outBake-outBake-outBake-out method methodmethod method

InflowInflowInflow Boundary Boundary Boundary N/A UUmeanUmeanmean = =0.324 0.324 0.324 m/s, m/s, m/s, N/AN/A UUUmeanmeanmean = =0.054= 0.054 0.054 m/s, m/s, m/s, InflowInflowInflow Boundary Boundary Boundary −1 UUmaxUmaxmax = =0.53 0.53 0.53 m/s, m/s, m/s, Air change rate = 0.5 h−1−1 −1 AirAir change change rate rate = =0.5 0.5 h h −1 AirAirAir changeUchangeUU meanchangemeanmean = = =0.324 rate 0.324 rate0.324 rate = =m/s, m/s, 3.0=3.0m/s, 3.0 hh −h 1 N/AN/AN/A UUUmeanmeanmean = = =0.054 0.054 0.054 m/s, m/s, m/s, U = outflow UUUmaxmaxmax = == =0.53 outflow0.53 0.53 m/s, m/s, m/s, out −1−1−1 out Outflow Boundary AirAirAir change change UchangeUoutout = = outflowrate rate outflowrate = = =0.5 0.5 0.5 h h h UUoutout = =outflow outflow −1−1−1 OutflowOutflow Boundary Boundary (Mass flow conservation) (MassAirAirAir change change change flow conservation)rate rate rate = = =3.0 3.0 3.0 h h h (Mass(Mass flow flow conservation) conservation) (Mass(Mass flow flow conservation) conservation) Installation of SBM N/A N/A Ceiling (A = 3.24 m2) InstallationInstallation of of SBM SBM UUUoutoutout = N/A= =outflow N/Aoutflow outflow UUUoutoutout = N/A= =outflow N/Aoutflow outflow Ceiling Ceiling (A (A = =3.24 3.24 m m2) 2) OutflowOutflowOutflowOperating Boundary Boundary Boundary time 2 h 2 h 2 h OperatingOperating time time (Mass(Mass(Mass flow flow flow 2conservation) conservation) 2hconservation) h (Mass(Mass(Mass flow flow flow 2conservation) conservation) 2hconservation) h 2 2h h Energy consumption N/A energy = 50 W N/A EnergyInstallationEnergyInstallationInstallation consumption consumption of of of SBM SBM SBM N/A N/A N/A Fan Fan energy energy N/A N/A N/A = =50 50 W W Ceiling Ceiling Ceiling (A N/A(A (A N/A = = = 3.24 3.24 3.24 m m m2)2 )2 ) ε TurbulentTurbulentOperatingOperatingTurbulentOperating model model timemodeltime time 2 2 2h h h Low Low Low Reynolds Reynolds Reynolds number number number 2k- 2 2hk- εh k- h εturbulence turbulenceturbulence model model model 2 2 2h h h Number of meshes Around 400,000 EnergyEnergyNumberEnergyNumber consumption consumption consumption of of meshes meshes N/A N/A N/A Around Fan AroundFan Fan energy energy energy 400,000 400,000 = = =50 50 50 W W W N/A N/A N/A Wall boundary No-slip TurbulentTurbulentWallTurbulentWall boundary boundary model model model Low Low Low Reynolds Reynolds Reynolds number number number No-slip No-slip k- k- k-εε εturbulence turbulence turbulence model model model Wall temperature The experimental result was set as the value of each wall surface temp. (Figure4) NumberNumberNumber of of of meshes meshes meshes Around Around Around 400,000 400,000 400,000 WallWallPollutant temperature temperature The The experimental experimental result result Toluene was was (Molecularset set as as the the value value weight, of of each 92.13each wall wall kg/kgmol) surface surface temp. temp. (Figure (Figure 4) 4) WallWallTimeWallPollutantPollutant boundary boundary stepboundary size Toluene Toluene (Molecular (Molecular No-slip No-slip No-slipweight, weight, 30 s 92.13 92.13 kg/kgmol) kg/kgmol) WallWallWallTimeTime temperature temperature temperature step step size size The The The experimental experimental experimental result result result was was was set set set as as as the the the value30 value 30value s s of of of each each each wall wall wall surface surface surface temp. temp. temp. (Figure (Figure (Figure 4) 4) 4) SBM: Sorptive building material. PollutantPollutantPollutant SBM:SBM: Sorptive Toluene TolueneSorptive Toluene (Molecularbuilding (Molecular (Molecularbuilding material. material. weight, weight, weight, 92.13 92.13 92.13 kg/kgmol) kg/kgmol) kg/kgmol) TimeTimeTime step step step size size size 30 30 30 s s s InIn the the case case of of natural natural ventilation ventilationSBM:SBM:SBM: and andSorptive Sorptive Sorptive mechanical mechanical building building building ventilation, ventilation,material. material. material. the the inlet inlet boundary boundary conditions conditions of of In the case of natural ventilation and mechanical ventilation, the inlet boundary conditions of the thethe airflow airflow were were calculated calculated to to correspond correspond to to air air changes changes per per hour hour (ACH) (ACH) of of 0.5 0.5 h −h1 −1and and 3 3h −h1,− 1, airflow were calculated to correspond to (ACH) of 0.5 h−1 and 3 h−1, respectively. respectively.respectively.InInIn thethe the casecase caseA A separate of of separateof naturalnatural natural velocity velocity ventilationventilation ventilation profile profile andand and was was mechanicalmechanical mechanical written written for ventilation,forventilation, ventilation, mechanical mechanical thethe the ventilation inletventilationinlet inlet boundaryboundary boundary and and the conditionstheconditions conditions maximum maximum ofof of A separate velocity profile was written for mechanical ventilation and the maximum wind−1−1−1 velocity−1−1−1 thewindthethewind airflow airflow airflowvelocity velocity were werewere was was calculated calculatedsetcalculated set to to 0.53 0.53 to totom/s. m/s.correspond correspondcorrespond Natural Natural ventto toventto air airilationair ilation changes changeschanges and and per SBMper perSBM hour hour hourinstallation installation (ACH) (ACH)(ACH) ofmethods ofofmethods 0.5 0.50.5 h hh are and areandand passive passive 3 3 3 h hh, , , was set to 0.53 m/s. Natural ventilation and SBM installation methods are passive methods and do respectively.methodsrespectively.respectively.methods and and A AdoA separatedo separateseparate not not consume consume velocity velocityvelocity energy. profileenergy. profileprofile wasFurthermore, waswasFurthermore, written writtenwritten for forforthe the mechanical mechanical mechanicalSBM SBM installation installation ventilation ventilationventilation method method and andand the thehas thehas maximum maximummaximumexcellent excellent not consume energy. Furthermore, the SBM installation method has excellent concentration reduction windconcentrationwindwindconcentration velocity velocityvelocity reduction was wasreductionwas set setset to toperformanceto performance0.53 0.530.53 m/s. m/s.m/s. Natural Natural Naturalagainst against vent ventvolatilevent volatileilationilationilation pollutants, pollutants, and andand SBM SBMSBM such suchinstallation installationinstallation as as toluene, toluene, methods methodsmethods without without are areare incurring incurring passive passivepassive performance against volatile pollutants, such as toluene, without incurring any additional operating methodsanymethodsmethodsany additional additional and andand do dooperatingdo operating not notnot consume consumeconsume cost. cost. Toluene, energy.Toluene, energy.energy. whichFurthermore, Furthermore,whichFurthermore, wa was sfound found the thethe inSBM SBMinSBM the the installation installationhighestinstallation highest concentration concentration method methodmethod has hashas during during excellent excellentexcellent the the cost. Toluene, which was found in the highest concentration during the experiment, was used as the concentrationexperiment,concentrationconcentrationexperiment, was wasreduction reduction reduction used used as as performancetheperformance performancethe pollutant. pollutant. against against Theagainst The analys analys volatile volatile volatileisis was waspollutants, pollutants, pollutants, conducted conducted such such such for for as 2as as 2 htoluene, toluene, htoluene,in in the the unsteady without unsteadywithout without incurring state.incurring incurringstate. anypollutant.anyany additional additionaladditional The analysis operating operatingoperating was cost. cost.cost. conducted Toluene, Toluene,Toluene, for which whichwhich 2 h in wa thewawass unsteadys found foundfound in inin state. the thethe highest highesthighest concentration concentrationconcentration during duringduring the thethe experiment,experiment,experiment, was was was used used used as as as the the the pollutant. pollutant. pollutant. The The The analys analys analysisisis was was was conducted conducted conducted for for for 2 2 2h h hin in in the the the unsteady unsteady unsteady state. state. state. Int. J. Environ. Res. Public Health 2018, 15, x 6 of 16 interpretations were performed in an unsteady state to observe the changes in pollutant concentration over time.

Table 4. Boundary condition of CFD (computational fluid dynamics) analysis for bake-out.

Classification ALT 1 ALT 2 ALT 3 Natural ventilation Mechanical ventilation Installation of SBM

Bake-out method

Inflow Boundary Umean = 0.324 m/s, N/A Umean = 0.054 m/s, Umax = 0.53 m/s, −1 Air change rate = 0.5 h Air change rate = 3.0 h−1 Uout = outflow Uout = outflow Outflow Boundary (Mass flow conservation) (Mass flow conservation) Installation of SBM N/A N/A Ceiling (A = 3.24 m2) Operating time 2 h 2 h 2 h Energy consumption N/A Fan energy = 50 W N/A Turbulent model Low Reynolds number k-ε turbulence model Number of meshes Around 400,000 Wall boundary No-slip Wall temperature The experimental result was set as the value of each wall surface temp. (Figure 4) Pollutant Toluene (Molecular weight, 92.13 kg/kgmol) Time step size 30 s SBM: Sorptive building material.

In the case of natural ventilation and mechanical ventilation, the inlet boundary conditions of the airflow were calculated to correspond to air changes per hour (ACH) of 0.5 h−1 and 3 h−1, respectively. A separate velocity profile was written for mechanical ventilation and the maximum wind velocity was set to 0.53 m/s. Natural ventilation and SBM installation methods are passive methods and do not consume energy. Furthermore, the SBM installation method has excellent concentration reduction performance against volatile pollutants, such as toluene, without incurring Int.any J. Environ.additional Res. Publicoperating Health 2018cost., 15 Toluene,, 2720 which was found in the highest concentration during 7 of the 16 experiment, was used as the pollutant. The analysis was conducted for 2 h in the unsteady state.

Int. J. Environ. Res. Public Health 2018, 15, x 7 of 16 (a)

(b)

(c)

FigureFigure 4.4. TemperatureTemperaturechange change over over time time per per case: case: ( a()a Case) Case 1 1 (Floor (Floor heating heating set set temp. temp. 30 30◦C); °C); (b ()b Case) Case 2 ◦ ◦ (Floor2 (Floor heating heating set set temp. temp. 40 40C); °C); (c )(c Case) Case 3 (Floor3 (Floor heating heating set set temp. temp. 50 50C). °C).

2.2.2. Turbulence Model and Governing Equations for CFD Analysis 2.2.2. Turbulence Model and Governing Equations for CFD Analysis According to details from existing studies, the SST k-ω model can predict the turbulence accurately According to details from existing studies, the SST k-ω model can predict the turbulence under the advanced wall function condition when the Y + value is less than 10 [21]. On the other hand, accurately under the advanced wall function condition when the Y + value is less than 10 [21]. On when the Y + value is less than 300, the standard k-e model can be applied to improve the convenience the other hand, when the Y + value is less than 300, the standard k-e model can be applied to of input parameters and lattice configuration. However, in this case, the inaccuracies in flow prediction improve the convenience of input parameters and lattice configuration. However, in this case, the at the separation, swirling flow and nearest wall should be considered. Thus, we used a low Reynolds inaccuracies in flow prediction at the separation, swirling flow and nearest wall should be number k-ε model instead of the standard k-ε model, which is known to accurately predict the natural considered. Thus, we used a low Reynolds number k-ε model instead of the standard k-ε model, due to buoyancy and small changes in the near-wall turbulent energy [22]. The fluid in the which is known to accurately predict the natural convection due to buoyancy and small changes in the near-wall turbulent energy [22]. The fluid in the space was set as an incompressible ideal gas and the buoyancy was set to be generated according to the density change. The continuity, momentum and energy equations are shown in Equations (2)–(4).

∂ρ + ∇ ∙ρu = 0 (2) ∂t

∂(ρu) + ∇ ∙ρuu = − p + ρg + ∇ ∙ (µ∇u) − ∇ ∙ τ (3) ∂t t

∂(ρe) + ∇ ∙ρeu = ∇ ∙ (k ∇T) − ∇ ∙ ( h j ) ∂t eff i i (4) i where ρ is the density of the fluid, t is the time, u is the fluid velocity vector, p is the pressure, g is the vector of gravitational acceleration, µ is the molecular dynamic viscosity, e is the specific internal energy, keff is the effective heat conductivity, T is the fluid temperature, hi is the specific of Int. J. Environ. Res. Public Health 2018, 15, 2720 8 of 16 space was set as an incompressible ideal gas and the buoyancy was set to be generated according to the density change. The continuity, momentum and energy equations are shown in Equations (2)–(4).

∂ρ + ∇·(ρu) = 0 (2) ∂t

∂(ρu) + ∇·(ρuu) = − ∇p + ρg + ∇·(µ ∇ u) − ∇·τ (3) ∂t t ! ∂(ρe) + ∇·(ρeu) = ∇·(k e f f ∇ T) − ∇· ∑ hi ji (4) ∂t i where ρ is the density of the fluid, t is the time, u is the fluid velocity vector, p is the pressure, g is the vector of gravitational acceleration, µ is the molecular dynamic viscosity, e is the specific internal energy, keff is the effective heat conductivity, T is the fluid temperature, hi is the specific enthalpy of the fluid and ji is the mass flux of the i-th constituent. The last term on the right hand side of Equation (3) is the divergence of the turbulence stresses (Reynolds stresses) and τt, accounts for the auxiliary stresses due to velocity fluctuations.

2.2.3. Pollutant Diffusion Model This study employed a species transport model for the dissemination of pollutants. The atmosphere inside the test space comprised air gas, simulating fresh outdoor air and toluene. Each constituent gas was set as an incompressible ideal gas with a temperature dependency. IDs were given to the emitted pollutants to enable quantitative measurement of the changes in their concentration. The partial differential transport equation, which was used to measure the pollutant concentration when the ID-ed pollutant passed through the Control Volume (CV) of the three-dimensional space, was derived from Equation (5) [23].

∂Y i + ∇·(ρY u) = − ∇·j (5) ∂t i i where Yi is the mass fraction of the i-th air constituent. Due to the reasonably low thermal parameters (pressure and temperature) of air, it can be treated as a rarified mixture. Hence, the mass flux of the i-th constituent may be calculated using Equation (6).

ji = − De f f ∇·Yi (6) where Deff is the effective diffusion coefficient, which includes the turbulence effects. The pollutant concentrations were calculated under the premise that the target pollutants are passive. Passive pollutant analysis is a simple method for calculating the pollutant distribution. It assumes that the pollutant and air share the same properties and generally defines the Schmidt number (=v/Da) as 1.0. However, the diffusion coefficient Da should be considered because it is an important boundary condition that affects the diffusion of the pollutant concentration in CFD analysis. In general, Hirschfelder equation is widely used to calculate the air diffusion coefficient; however, it is often inconsistent with the experimental value. In this study, the Fujita equation is used to derive and apply the diffusion coefficient of toluene under different temperatures. The diffusion of toluene and the analysis of sorption were performed and toluene was assumed to be a passive contaminant. The concentration distribution was produced and the water (vapor) diffusion in air was expressed as shown in Equation (7). The water (vapor) diffusion coefficient in air was calculated using Equations (8)–(10) [24,25].

  ! ∂C1 ∂UiC1 ∂ vt ∂C1 + = Da + (7) ∂t ∂xi ∂xi Sct ∂xi Int. J. Environ. Res. Public Health 2018, 15, 2720 9 of 16

A − B log P = − 3 (8) 10 w C + T

M1 Pw C0 = ρa × × (9) M2 P − Pw −3 −8 1.83 " 1/3  1/3# s 6.7 × 10 ×T Tc1 Tc2 1 1 Da = × + + (10) P Pc1 Pc2 M1 M2

3 where C1 is the pollutant concentration at a spatial point (µg/m ), Da is the molecular pollutant 2 2 diffusion coefficient (m /s), Ui is the wind velocity (m/s), vt is the eddy viscosity (m /s), Sct is the turbulent Schmidt’s number (-) and Pw is the water vapor pressure (Pa). In addition, A, B and C are the empirical constants with values of 7.7423, 1554.16 and 219, respectively. T denotes the temperature ◦ 3 3 ( C), C0 is the saturation concentration (g/m ) and ρa is the air density (g/m ). M1 and M2 are the molecular weights, P is the atmospheric pressure in the chamber (Pa), Tc1 and Tc2 are the critical ◦ temperatures ( C) and Pc1 and Pc2 are the critical pressure values (Pa).

2.2.4. Pollutant Adsorption Model This study used CFD analysis to assess the reduction in contaminant concentration when SBM is used in the bake-out process. The pollutant adsorption model was examined and applied to the CFD analysis. The adsorption of contaminants was measured under the premise that the surface density Cs of SBM is zero. This is consistent with the Henry constant Kh = ∞ in the Henry adsorption isotherm, which is used when the adsorbent in questions has an outstanding adsorption effect. In this study, the adsorbent flux adsorption model for the surface of SBM was applied to a single material. The adsorption model was applied to homogeneous materials and the adsorption flux, or fluxad, on SMB surfaces was found using Equation (11) [26].

∂C CCV − Ceq f luxad = −De f f = −De f f (11) ∂x B ∆x where CCV is the contaminant concentration of the first control volume (CV) of air coming in contact 3 0 3 with the SBM surface (µg/m ), Ceq is the SBM surface s adsorption concentration (µg/m ) and Deff is the effective diffusion coefficient (m2/s).

3. Results and Discussion

3.1. Floor Temperature Control and Energy Consumption Figure4 shows the temperature change in the sensors for each case over time. Heating was turned off during the first 30 min of the experiment and Cases 1–3 showed similar temperature changes when cooling by airing was administered under identical conditions. As the results show, a dramatic change in temperature was detected once heating was turned on and the floor temperatures reached steady states in relation to the set values within 30 min in all cases. This is typical of PID control that has a high initial response speed. In Case 1, where the floor temperature was set to 30 ◦C, both the maximum and minimum temperatures of all sensors were within ±0.8 ◦C from the set values with only small differences. The average temperature was measured to be identical to the set value at 30 ◦C and the standard deviation and range were found to be similar across all sensors. This situation occurs owing to the target value0s high proximity to the set value as a result of PID control as well as the presence of very little offset. On the other hand, the temperature distribution of Case 2 shows a slight increase in the range of oscillation with respect to the measured temperature based on the set value. However, considering that the peak load of a typical on/off control is ±4 ◦C, floor heating by PID control and PTC film was deemed reliable for maintaining the set temperature. Int. J. Environ. Res. Public Health 2018, 15, 2720 10 of 16

Figure5 shows the energy and power consumption over time per case. Maximum instantaneous energy consumption was observed immediately after heating was turned on and its oscillation range subsequently lowered as the target value approached the set value. In Case 1, the initial manipulated variable did not show dramatic change in relation to the set time for the derivative control of PID. For Case 3, on the other hand, a substantial change in the manipulated variable was seen during the set time and maximum energy was continuously supplied for about 1 h after the heating was turned on. The energy consumption for Case 1 with floor surface temperature set at 30 ◦C was 974 Wh. On the other hand, the total energy consumption in Case 2, whose difference in floor temperature was 13 ◦C Int.before J. Environ. and afterRes. Public the heating,Health 2018 was, 15, x 3634 Wh, which is about 2.7 times greater than that of Case 1.10 of 16

Int. J. Environ. Res. Public Health 2018, 15, x 10 of 16

Figure 5. Energy and power consumption from heating under PID (proportional integral derivative) derivative) controlFigure each 5. Case. Energy and power consumption from heating under PID (proportional integral derivative) control each Case. 3.2. Quality Assurance of Pollutant Measurement and Analysis 3.2. Quality Assurance of Pollutant Measurement and Analysis 3.2. Quality Assurance of Pollutant Measurement and Analysis This study validated pollutant measurement and analysis by assessing linearity, precision and This study validated pollutant measurement and analysis by assessing linearity, precision and accuracy perThis thestudy ICH validated Guidelines pollutant [27]. measurement and analysis by assessing linearity, precision and accuracy per the ICH Guidelines [27]. Linearityaccuracy per is relatedthe ICH toGuidelines the accuracy [27]. and scope of the analysis and in this study plots of signal Linearity is related to the accuracy and scope of the analysis and in this study plots of signal strength againstLinearity concentrations is related to the ofaccuracy VOCs wereand scope visually of the evaluated. analysis and When in this a study linear plots relationship of signal was strengthstrength against against concentrations concentrations of ofVOCs VOCs were were visuvisually evaluated.evaluated. When When a lineara linear relationship relationship was was recognized, the measurement method was assessed using statistical methods, such as calculating the recognized,recognized, the measurementthe measurement method method was was assessed assessed using statisticalstatistical methods, methods, such such as calculating as calculating the the slope of the line using the least squares method. A total of six VOC concentrations were used for the slopeslope of the of line the lineusing using the the least least squares squares method. method. AA total ofof six six VOC VOC concentrations concentrations were were used used for the for the study: 5 ppb, 10 ppb, 30 ppb, 100 ppb, 300 ppb and 1000 ppb. study:study: 5 ppb, 5 ppb, 10 ppb, 10 ppb, 30 ppb,30 ppb, 100 100 ppb, ppb, 300 300 ppb ppb and and 1000 ppb.ppb. FigureFigure6 shows 6 shows the linear the linear calibration calibration curves curves of each of each VOC VOC fit according fit according to the to leastthe least squares squares method. Figure 6 shows the linear calibration curves of each VOC fit according to the least squares The correlationmethod. The coefficient correlation (R 2coefficient) was greater (R2) thanwas greater 0.99 for than all of 0.99 them. for Hear,all of thethem. relation Hear, shouldthe relation be linear method. The correlation coefficient (R2) was greater than 0.99 for all of them. Hear, the relation in theshould log-log be scale linear because in the log-log of semiconductor scale because of characteristics. semiconductor characteristics. should be linear in the log-log scale because of semiconductor characteristics.

FigureFigure 6. Calibration 6. Calibration curve curve of of VOCs VOCs (Toluene,(Toluene, Ethylbenzene, m-Xylene m-Xylene and and Styrene). Styrene).

To verify precision, retention time and signal strength were compared after measuring VOCs Figure 6. Calibration curve of VOCs (Toluene, Ethylbenzene, m-Xylene and Styrene). with identical gas concentrations eight times. Table 5 shows the mean signal strength and concentration of toluene, ethylbenzene, m-xylene and styrene. The relative standard deviations To verify precision, retention time and signal strength were compared after measuring VOCs (RSD) of the VOCs were moderate, ranging from 2.07 to 2.82%. with identicalAccuracy gas was concentrations measured by repeatedly eight times. measuring Table fo ur5 differentshows theconcentrations mean signal four timesstrength each and concentration(Table 6). Theof toluene,accuracy ofethylbenzene, the VOCs concentrat m-xyleneion measurementand styrene. was The between relative 98.00% standard and 112.11%, deviations (RSD)with of the the VOCs relative were standard moderate, deviation ranging (RSD) from < 2.84%.2.07 to Also, 2.82%. the RSD of measured concentration Accuracyranging from was 10 measured ppb to 100 byppb repeatedly was lower thanmeasuring 1.02%. four different concentrations four times each (Table 6). The accuracy of the VOCs concentration measurement was between 98.00% and 112.11%, with the relative standard deviation (RSD) < 2.84%. Also, the RSD of measured concentration ranging from 10 ppb to 100 ppb was lower than 1.02%. Int. J. Environ. Res. Public Health 2018, 15, 2720 11 of 16

To verify precision, retention time and signal strength were compared after measuring VOCs with identical gas concentrations eight times. Table5 shows the mean signal strength and concentration of toluene, ethylbenzene, m-xylene and styrene. The relative standard deviations (RSD) of the VOCs were moderate, ranging from 2.07 to 2.82%.

Table 5. Comparison of VOCs concentrations for precision.

Compound Retention Time (s) Signal Strength (mV) Measured Concentration (ppb) RSD (%) Toluene 92.16 ± 1.57 (1) 154.42 ± 3.21 103.77 ± 2.96 2.07 Ethylbenzene 171.78 ± 1.44 132.03 ± 2.97 103.96 ± 3.01 2.25 m-Xylene 211.12 ± 0.95 135.80 ± 3.22 102.90 ± 3.39 2.37 Styrene 307.16 ± 1.78 72.92 ± 2.05 104.40 ± 3.99 2.82 (1) Values are presented as mean ± standard deviation (n = 8).

Accuracy was measured by repeatedly measuring four different concentrations four times each (Table6). The accuracy of the VOCs concentration measurement was between 98.00% and 112.11%, with the relative standard deviation (RSD) < 2.84%. Also, the RSD of measured concentration ranging from 10 ppb to 100 ppb was lower than 1.02%.

Table 6. Comparison of VOCs concentrations for accuracy.

Injection Concentration Signal Strength Measured Compound Accuracy (%) RSD (%) (ppb) (mV) Concentration (ppb) 10 53.49 ± 0.46 (1) 10.67 ± 0.42 102.11–112.10 0.86 30 76.85 ± 0.82 32.22 ± 0.75 104.10–111.00 1.07 Toluene 100 158.95 ± 3.48 107.95 ± 3.21 104.80–111.20 2.19 300 377.18 ± 3.66 309.25 ± 3.38 101.60–104.10 0.97 10 39.55 ± 0.30 10.37 ± 0.30 102.00–108.00 0.77 30 60.18 ± 0.79 31.25 ± 0.81 102.00–108.20 0.79 Ethylbenzene 100 133.10 ± 1.08 105.05 ± 1.09 103.60–105.90 0.10 300 332.11 ± 2.84 306.45 ± 2.88 100.73–102.83 2.84 10 47.82 ± 0.32 10.25 ± 0.34 98.00–105.90 0.67 30 67.33 ± 0.34 30.80 ± 0.36 101.33–104.10 0.51 m-Xylene 100 139.70 ± 3.25 107.02 ± 3.42 102.40–110.00 2.32 300 325.01 ± 1.65 302.15 ± 1.74 100.00–101.40 0.51 10 24.70 ± 0.08 10.82 ± 0.17 108.00–110.00 0.35 30 35.32 ± 0.43 31.42 ± 0.85 100.67–106.67 1.23 Styrene 100 74.96 ± 0.91 108.35 ± 1.78 106.80–110.90 1.22 300 178.69 ± 1.46 309.65 ± 2.84 101.87–104.07 0.82 (1) Values are presented as mean ± standard deviation (n = 4).

3.3. Change in Pollutant Concentration According to Floor Temperature (Bake Process) Table7 shows the results of 10 measurements of background concentration in the test chamber prior to the injection of pollutant. In Case 3, where the floor temperature was set to 50 ◦C, more contaminants were emitted compared to the other cases. Thus, it can be deduced that residual contaminants were discharged as the test body was heated. Five pollutant types were detected during the measurement period, in which o-xylene was not included. Further, toluene concentration appeared to be the highest in all cases.

Table 7. Background concentration of test chamber.

Case Set Temp. (◦C) Toluene (ppb) Ethylbenzene (ppb) m-Xylene (ppb) o-Xylene (ppb) Styrene (ppb) Case 1 30 17.2 4.1 3.8 - 2.6 Case 2 40 18.4 3.7 5.6 - 4.3 Case 3 50 18.7 4.4 5.3 - 2.8

Figure7a–c show the changes in contaminant concentrations per case. Figure7d illustrates the changes in concentration of toluene and the results of the CFD analysis. It was found that Int. J. Environ. Res. Public Health 2018, 15, x 12 of 16

Int. J. Environ. Res. Public Health 2018, 15, 2720 12 of 16 against the side walls. The boundary layer of toluene concentration also disseminated at the center of the floor. However, considering that the average wind velocity in the chamber is 0.02 m/s, it can behigher concluded floor temperatures that rise in leadtemperature to higher has emission more effect rates acrosson toluene all pollutant diffusion types than but the m-xylene natural ventilation.to a lesser degree than others. Thus, measures other than increasing the surface temperature are necessaryTable to8 reduce shows the the pollutant relationship concentration between of m-xylene, energy whichconsumption is emitted and from rate building of pollutant materials. concentrationThe maximum increase concentrations per case. of tolueneWhen the detected floor heating in the center temperature of the test was chamber set to 50 during °C, the the amount pollutant of energymeasurement consumed period increased were 41.0 to ppb,about 59.4 five ppb times and as 80.3 that ppb of Case for Case 1 but 1, the Case maximum 2 and Case concentration 3, respectively. of tolueneExperimentation detected increased and CFD analysisby a factor ofCase of approxim 3 showedately that 0.9. the This differences shows inthat toluene although concentrations raising the weretemperature within aduring 5% margin the bake-out of error, process lending helps credence to increase to CFD pollutant analysis basedemission, on thethe pollutionconcentrations diffusion are notmodel proportionally employed in increased. this study. Hence, a more efficient and economical operation method is needed.

(a) (b)

(c) (d)

Figure 7. ConcentrationsConcentrations of of pollutants pollutants per per case: case: (a (a) )Case Case 1; 1; ( (bb)) Case Case 2; 2; ( (cc)) Case Case 3; 3; ( (dd)) Toluene concentration of each case. CFD: computational fluid fluid dynamics.

The CFD analysis of Case 3 at time (600 s) is illustrated in Figure8. The ∆T value between the floor and side walls were observed to be 25 ◦C, which resulted in natural convection within the test chamber. The air current ascended at the center of the floor, then cooled down and descended against the side walls. The boundary layer of toluene concentration also disseminated at the center of the floor. However, considering that the average wind velocity in the chamber is 0.02 m/s, it can be concluded that rise in temperature has more effect on toluene diffusion than the natural ventilation. Table8 shows the relationship between energy consumption and rate of pollutant concentration increase per case. When the floor heating temperature was set to 50 ◦C, the amount of energy consumed increased to about five times as that of Case 1 but the maximum concentration of toluene detected increased by a factor of approximately 0.9. This shows that although raising the temperature during the bake-out process helps to increase pollutant emission, the concentrations are not proportionally increased.Figure Hence,8. Results a moreof CFD efficient (computational and economical fluid dynamics) operation analysis method (Case is 3, needed.time = 600 s, Section XY, Z = 0.9 m). Int. J. Environ. Res. Public Health 2018, 15, x 12 of 16 against the side walls. The boundary layer of toluene concentration also disseminated at the center of the floor. However, considering that the average wind velocity in the chamber is 0.02 m/s, it can be concluded that rise in temperature has more effect on toluene diffusion than the natural ventilation. Table 8 shows the relationship between energy consumption and rate of pollutant concentration increase per case. When the floor heating temperature was set to 50 °C, the amount of energy consumed increased to about five times as that of Case 1 but the maximum concentration of toluene detected increased by a factor of approximately 0.9. This shows that although raising the temperature during the bake-out process helps to increase pollutant emission, the concentrations are not proportionally increased. Hence, a more efficient and economical operation method is needed.

(a) (b)

(c) (d)

Figure 7. Concentrations of pollutants per case: (a) Case 1; (b) Case 2; (c) Case 3; (d) Toluene Int. J. Environ. Res. Public Health 2018, 15, 2720 13 of 16 concentration of each case. CFD: computational fluid dynamics.

Int. J. Environ.Figure 8. Res. ResultsResults Public Healthof of CFD CFD 2018 (computational (computational, 15, x fluid fluid dynamics) dynamics) analysis analysis (Case (Case 3, 3,time time = 600 = 600 s, Section s, Section XY, XY, Z13 of 16 =Z 0.9 = 0.9 m). m). Table 8. Energy consumption and maximum concentration of toluene per case. Table 8. Energy consumption and maximum concentration of toluene per case. Case Et (Rate of Increase) Em (Rate of Increase) Maximum Concentration of Toluene (Rate of Increase) CaseCase 1 E 974t (Rate Wh of(-) Increase) Em 852(Rate Wh of (-) Increase) Maximum Concentration 41.0 ofppb Toluene (-) (Rate of Increase) CaseCase 2 1 3634 Wh 974 (273%) Wh (-) 2755 Wh 852 (223%) Wh (-) 59.4 41.0 ppb ppb (45%) (-) CaseCase 3 2 6789 3634Wh (597%) Wh (273%) 5280 2755 Wh Wh (519%) (223%) 80.3 59.4 ppb ppb (45%)(96%) Case 3 6789 Wh (597%) 5280 Wh (519%) 80.3 ppb (96%) Et: Total electric energy consumption including the stabilization period. Em: Electric energy Et: Total electric energy consumption including the stabilization period. Em: Electric energy consumption during consumptionpollutant measurement during polluta period. nt measurement period.

3.4. Evaluation of Contaminant Concentration Reduction through Use of Sorptive Building Material and 3.4. Evaluation of Contaminant Concentration Reduction through Use of Sorptive Building Material and Ventilation (Bake-out Process)Process) Figure 99 showsshows thethe changeschanges inin toluenetoluene concentrationconcentration perper ALTALT during during bake-out. bake-out. TheThe toluenetoluene concentration in the chamber continuously increased increased for for 8 h during floor floor heating and immediately after the bake-out started, the toluene concentratio concentrationn decreased rapidly. As a result, very low toluene concentrations were observed after 2 h regardless ofof thethe bake-outbake-out method.method.

Figure 9. Figure 9. ConcentrationConcentration of of pollutant pollutant during during Bake-o Bake-outut process process using CFD (computational fluid fluid dynamics) analysis. SBM: sorptive building material. dynamics) analysis. SBM: sorptive building material. Table9 shows the results of the step-down tracer gas test to compare the energy consumption Table 9 shows the results of the step-down tracer gas test to compare the energy consumption (Em with fan energy), age-of-air and energy consumption of each ALT that occurred during the local (Em with fan energy), age-of-air and energy consumption of each ALT that occurred during the local age-of-air and bake-out processes. Local age-of-air is a measure proposed by ASHRAE Standard 129 to evaluate the ventilation efficiency of a target space and can be calculated using Equation (12) [28].

Ai = ∇τ ∙ Ci,avg/Ci,start (12) where τ is time period of tracer gas measurement, Ci,avg is the average tracer gas concentration at location i during ∇τ and Ci,start is the initial tracer gas concentration at location i. Equation (12) enables calculation of the local age-of-air by defining the entire space as a single CV in the CFD analysis. Even if the target space is airtight, reduction in the concentration of indoor air pollutant due to SBM installation and air filtering can be interpreted as synonymous of a ventilation effect. In such a case, a low age-of-air value implies high effectiveness in reducing contamination, similar to ventilation. In contrast, this study calculated the age-of-air by allocating the center of the test chamber, which is in the same position as in the experiment, as local. The age-of-air was 1605 s when bake-out was performed via mechanical ventilation. Further, it was 1936 s with installation of SBM (ALT 3) Int. J. Environ. Res. Public Health 2018, 15, 2720 14 of 16 age-of-air and bake-out processes. Local age-of-air is a measure proposed by ASHRAE Standard 129 to evaluate the ventilation efficiency of a target space and can be calculated using Equation (12) [28].

Ai = ∇τ·Ci,avg/Ci,start (12) where ∇τ is time period of tracer gas measurement, Ci,avg is the average tracer gas concentration at location i during ∇τ and Ci,start is the initial tracer gas concentration at location i.

Table 9. Pollutant Concentration Reduction Effect per alternative (ALT).

Decreasing Rate of Concentration Per Energy ALT E with Fan Energy (Wh) Age-of-Air (s) m Consumption (ppb/kWh) ALT 1 5280 2156 10.68 ALT 2 5380 1632 11.34 ALT 3 5280 1936 11.30

Em: Electric energy consumption during pollutant measurement period.

Equation (12) enables calculation of the local age-of-air by defining the entire space as a single CV in the CFD analysis. Even if the target space is airtight, reduction in the concentration of indoor air pollutant due to SBM installation and air filtering can be interpreted as synonymous of a ventilation effect. In such a case, a low age-of-air value implies high effectiveness in reducing contamination, similar to ventilation. In contrast, this study calculated the age-of-air by allocating the center of the test chamber, which is in the same position as in the experiment, as local. The age-of-air was 1605 s when bake-out was performed via mechanical ventilation. Further, it was 1936 s with installation of SBM (ALT 3) and 2293 s with natural ventilation (ALT 1). Although ventilation was not performed with the installation of SBM, the SBM had a pollutant concentration reduction effect similar to ventilation as it adsorbed pollutants. This means that in an environment such as a closed space where natural ventilation is very poor or non-existent, the indoor pollutant concentration can be reduced by installing an SBM.

4. Conclusions This study tested the effectiveness of bake-out at reducing indoor VOC concentrations via a two-step process, using experimentation and CFD analysis. Floor heating was implemented using PID control so that pollutant reduction could be measured in relation to the amount of energy consumed. PID control is shown to be effective at controlling the temperature within a narrow range with a very small standard deviation, which provides a reliable boundary condition for steady state analysis and allows for experimental and CFD analyses to be performed simultaneously. A test chamber was created to analyze the change in contaminant concentration with floor temperature. Contaminant concentrations increased with the increase in floor temperatures but the induced emission rates were below expectation. These results were very similar to the results of previous studies. In this study, we also examined the adequacy of energy consumption in addition to the improvement of indoor air quality. In South Korea, where most households use floor heating systems, it would be more efficient to maintain a moderate temperature for an extended period of time rather than varying the temperatures. The present study observed how the bake-out process affected pollutant emissions in airtight spaces and used CFD analysis to analyze the performances of three ALTs. The pollutant diffusion models for CFD analysis were found to correspond closely to the experimental data. The mechanical ventilation system was the most effective at reducing pollutant concentration. Installing SBM was also found to be effective, which implies that in places where natural ventilation is unavailable or the ventilation amount is insignificant, SBM can be used for IAQ improvement. Furthermore, SBM does not require additional cost of operation and prevents heat loss due to inflow of outside air during winter and summer. Consequently, as consuming enormous heating energy to reduce pollutant concentration Int. J. Environ. Res. Public Health 2018, 15, 2720 15 of 16 can cause economic losses, economy and sustainability should be considered in every process of the bake-out. To reduce the pollution produced by household appliances and air-conditioning systems, new generation of low-consumption thermo-acoustic refrigerators, risk of polluting emissions and noise can be used [29]. However, the SBM used in this study was assumed to have infinite capacity and its performance may have been enhanced because the applied area was wider than the spatial volume. Moreover, it must be noted that ventilation is performed not to only remove pollutants; therefore, other factors that may require ventilation should be considered. Subsequent research will employ more variables, including ventilation methods, air changes, vent locations, location of SBM and its surface area and adsorbic capacity of SBM, to find the most efficient and economical method of pollutant reduction. In addition, this study focused on CFD analysis of the bake-out process. Therefore, we will carry out a further study based on experiments to verify the reliability of adsorption, removal and ventilation effects, including the diffusion of pollutants.

Author Contributions: S.P. provided main idea of the experiment and CFD analysis, analyzed the results of study and wrote the paper. J.S. reviewed the paper. All authors have read and approved the final manuscript. Funding: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. 2018R1A2B2007165). Conflicts of Interest: The authors declare no conflict of interest.

References

1. Barabad, M.; Jung, W.; Versoza, M.; Kim, M.; Ko, S.; Park, D.; Lee, K. Emission Characteristics of Particulate Matter, Volatile Organic Compounds, and Trace Elements from the Combustion of Coals in Mongolia. Int. Environ. Res. Public Health 2018, 15, 1706. [CrossRef][PubMed] 2. Salthammer, T.; Mentese, S.; Marutzky, R. Formaldehyde in the Indoor Environment. Chem. Rev. 2010, 110, 2536–2572. [CrossRef] 3. Smith, M.T. Advances in Understanding Benzene Health Effects and Susceptibility. Annu. Rev. Public Health 2010, 31, 133–148. [CrossRef][PubMed] 4. Du, Z.; Mo, J.; Zhang, Y. Risk assessment of population inhalation exposure to volatile organic compounds and carbonyls in urban China. Environ. Int. 2014, 73, 33–45. [CrossRef][PubMed] 5. Girman, J.R.; Alevantis, L.E.; Kulasingam, G.C.; Petreas, M.X.; Webber, L.M. The bake-out of an office building: A case study. Environ. Int. 1989, 15, 449–453. [CrossRef] 6. Girman, J.R. Volatile organic compounds and building bake-out. Occup. Med. 1989, 4, 695–712. [PubMed] 7. Lu, Y.; Liu, J.; Lu, B.-N.; Jiang, A.-X.; Zhang, J.-R.; Xie, B.; Li, J.; Zhang, L. Numerical value research on bake-out technology with dilution ventilation for building materials. Proc. Build. Simul. 2007, 2007, 907–911. 8. Edwards, R.D.; Jurvelin, J.; Koistinen, K.; Saarela, K.; Jantunen, M. VOC source identification from personal and residential indoor, outdoor and workplace microenvironment samples in EXPOLIS-Helsinki, Finland. Atmos. Environ. 2001, 35, 4829–4841. [CrossRef] 9. Kang, D.H.; Choi, D.H.; Lee, S.M.; Yeo, M.S.; Kim, K.W. Effect of bake-out on reducing VOC emissions and concentrations in a residential housing unit with a radiant floor heating system. Build. Environ. 2010, 45, 1816–1825. [CrossRef] 10. Lv, Y.; Liu, J.; Wei, S.; Wang, H. Experimental and simulation study on bake-out with dilution ventilation technology for building materials. J. Air Waste Manag. Assoc. 2016, 66, 1098–1108. [CrossRef][PubMed] 11. Lee, C.; Haghighat, F.; Ghaly, W. A study on VOC source and sink behavior in porous building materials-analytical model development and assessment. Indoor Air 2005, 15, 183–196. [CrossRef] 12. Jiang, C.; Li, D.; Zhang, P.; Li, J.; Wang, J.; Yu, J. Formaldehyde and volatile organic compound (VOC) emissions from particleboard: Identification of odorous compounds and effects of heat treatment. Build. Environ. 2017, 117, 118–126. [CrossRef] 13. Ataka, Y.; Kato, S.; Murakami, S.; Zhu, Q.; Ito, K.; Yokota, T. Study of effect of adsorptive building material on formaldehyde concentrations; development of measuring methods and modeling of adsorption phenomena. Indoor Air 2004, 14, 51–64. [CrossRef] 14. Seo, J.; Kato, S.; Ataka, Y.; Chino, S. Performance test for evaluating the reduction of VOCs in rooms and evaluating the lifetime of sorptive building materials. Build. Environ. 2009, 44, 207–215. [CrossRef] Int. J. Environ. Res. Public Health 2018, 15, 2720 16 of 16

15. Park, S.; Seo, J.; Kim, J.T. A study on the application of sorptive building materials to reduce the concentration and volume of contaminants inhaled by occupants in office areas. Energy Build. 2015, 98, 10–18. [CrossRef] 16. Graf, M.; Barrettino, D.; Taschini, S.; Hagleitner, C.; Hierlemann, A.; Baltes, H. Metal oxide-based monolithic complementary metal oxide semiconductor gas sensor microsystem. Anal. Chem. 2004, 76, 4437–4445. [CrossRef] 17. Fine, G.F.; Cavanagh, L.M.; Afonja, A.; Binions, R. Metal oxide semi-conductor gas sensors in environmental monitoring. Sensors 2010, 10, 5469–5502. [CrossRef] 18. Kim, S.; Choi, Y.-K.; Park, K.-W.; Kim, J.T. Test methods and reduction of organic pollutant compound emissions from wood-based building and furniture materials. Bioresour. Technol. 2010, 101, 6562–6568. [CrossRef] 19. Yang, X. Study of Building Material Emissions and Indoor Air Quality. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 1999. 20. ANSYS 17.2. Fluent Theory Guide; ANSYS Inc.: Canonsburg, PA, USA, 2011. 21. Hussain, S.; Oosthuizen, P.H.; Kalendar, A. Evaluation of various turbulence models for the prediction of the airflow and temperature distributions in atria. Energy Build. 2012, 48, 18–28. [CrossRef] 22. Abe, K.; Kondoh, T.; Nagano, Y. A new turbulence model for predicting fluid flow and in separating and reattaching flows—I. Flow field calculations. Int. J. Heat Mass Transf. 1994, 37, 139–151. [CrossRef] 23. Buli´nska,A.; Popiołek, Z.; Buli´nski,Z. Experimentally validated CFD analysis on sampling region determination of average indoor carbon dioxide concentration in occupied space. Build. Environ. 2014, 72, 319–331. [CrossRef] 24. The Society of Chemical Engineers. Handbook of Chemistry; The Society of Chemical Engineers: Tokyo, Japan, 1999. 25. Seo, J.; Lee, S.; Kim, S.; Nam, Y. Control of emission rates of chemical compounds emitted by controlling their mass transfer coefficients on the surface of the tested building material. J. Adhes. Sci. Technol. 2013, 27, 610–619. [CrossRef] 26. Lim, E.-S. Evaluation of Contaminant Concentration Reduction Effect of Sorptive Building Materials by New Ventilation Index-Net Escape Velocity. J. Archit. Inst. Korea Plan. Des. 2013, 29, 291–298. (In Korean) [CrossRef] 27. ICH Expert Working Group. Validation of analytical procedures: Text and methodology Q2(R1). In Proceedings of the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, Geneva, Switzerland, 25–27 April 2005. 28. ASHRAE. ASHRAE Standard 129—Measuring Air-Change Effectiveness; American Society of Heating, Refrigerating and Air-Conditioning Engineers: Atlanta, GA, USA, 2004. 29. Piccolo, A.; Siclari, R.; Rando, F.; Cannistraro, M. Comparative performance of thermoacoustic heat exchangers with different pore geometries in oscillatory flow. Implementation of experimental techniques. Appl. Sci. 2017, 7, 784. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).